如何增加子图文本大小并添加自定义条形图注释

Question

我有以下数据框，代表每个地区每个部门/区域的员工总数。

               Finance   HR   IT  Marketing  Medical  Operations  Sales
Business Unit                                                          
Cardiology           0    2    1          0        3           0      0
Genetics             1    4    3          1        3           1      1
Imaging             34   74   70         38       68          18     33
Labs                63  130  131         66      130          32     68
Pathology            2    5   10          4        8           3      6

使用此数据框，我使用此代码生成了波纹图：

#Plot the graph
fig, ax = plt.subplots(1, 5, figsize=(30, 15), sharey = True)
iax = iter(ax.flatten())
for n, g in df.groupby('Business Unit'):
    g.loc[n, :].plot.bar(ax=next(iax),  title=f'{n}', stacked=True, legend = True, fontsize = 30)

如您所见，通知业务部门的子图的文本大小以及图例大小（在顶角）太小了。我怎样才能增加他们两个的文本？

另外，有没有一种方法可以在这段代码中添加一种方法来显示每列中总数的百分比？

非常感谢您的帮助！

Answer 1

• 我会使用pandas.DataFrame.plot，它将索引绘制为xaxis。因此，数据帧需要用.T 转置
  • 这使用matplotlib 和默认的绘图后端
• 使用subplots=True，将df.plot 分配给axes，这是array 的arrays 的AxesSubplots
• 可以使用.bar_label添加注释，可以使用labels=参数自定义标签。以下链接包含自定义 labels 的示例。
  • Bar Label Demo
  • stack bar plot in matplotlib and add label to each section
  • How to add multiple annotations to a barplot
  • How to plot a horizontal stacked bar with annotations
• 在pandas 1.3.0 和matplotlib 3.4.2 中测试

import pandas as pd

# load dataframe
data = {'Cardiology': {'Finance': 0, 'HR': 2, 'IT': 1, 'Marketing': 0, 'Medical': 3, 'Operations': 0, 'Sales': 0}, 'Genetics': {'Finance': 1, 'HR': 4, 'IT': 3, 'Marketing': 1, 'Medical': 3, 'Operations': 1, 'Sales': 1}, 'Imaging': {'Finance': 34, 'HR': 74, 'IT': 70, 'Marketing': 38, 'Medical': 68, 'Operations': 18, 'Sales': 33}, 'Labs': {'Finance': 63, 'HR': 130, 'IT': 131, 'Marketing': 66, 'Medical': 130, 'Operations': 32, 'Sales': 68}, 'Pathology': {'Finance': 2, 'HR': 5, 'IT': 10, 'Marketing': 4, 'Medical': 8, 'Operations': 3, 'Sales': 6}}
df = pd.DataFrame.from_dict(data, orient='index')

# get the total for each business unit; used to calculate percent
bu_total = df.sum(axis=1)

# get the total for each division; used to calculate percent
div_total = df.sum(axis=0)

# plot
axes = df.T.plot(kind='bar', subplots=True, layout=(1, 5), figsize=(22, 6), sharey=True, ylabel='Expenditures ($)')

# iterate through axes subplots
for ax in axes[0]:
    # title is used to get the total from bu_total
    title = ax.get_title()
    
    ax.legend(fontsize=15)
    ax.set_xticklabels(ax.get_xticklabels(), fontdict={'fontsize':24})
    ax.set_title(title, fontdict={'fontsize':24})
    ax.set_ylabel(ax.get_ylabel(), fontdict={'fontsize':24})
    
    # customized labels for % business unit total
    bu_labels = [f'{(v.get_height() / bu_total[title])*100 :.0f}%' for v in ax.containers[0]]
    
    # customized labels for % division total
    div_labels = [f'{(v.get_height() / div_total[i])*100 :.0f}%' for i, v in enumerate(ax.containers[0])]
    
    # annotate as desired: use bu_lables or div_labels
    ax.bar_label(ax.containers[0], labels=bu_labels, label_type='edge', fontsize=10)

    # pad the spacing between the number and the edge of the figure
    ax.margins(y=0.1)